Transcript Mayors’ Reelection Choice and the Economy: Evidence from

Mayors’ Reelection Choice
and the Economy: Evidence
from Portugal
Vítor Castro and Rodrigo Martins
University of Coimbra, NIPE and GEMF, Portugal
EPCS 2011 Meeting
28th April – 1st May
Motivation

Lack of an analysis on the determinants of
Mayors’ decision of running for another term
in office: Economic? Political? Individual?

Fill that gap in the literature.

Empirical analysis for Portugal.

Portuguese local elections: 1979-2005.

Extensive data set: mainland municipalities.
2
Motivation

The Portuguese case => excellent laboratory:
1) election dates are defined exogenously from the
perspective of local authorities;
2) all municipalities have elections on the same day;
3) there are no legal limits on the number of terms in
office during the time span considered in this study;
4) local incumbents have a key influence in local
policy and outcomes.
3
Literature

Explores the influence of economic and
political conditions on the electoral results and
on politicians’ popularity [survey: Paldam(2004)]

and how voters judge democratic governments.

Explores the dynamics of political ambition
and politicians’ career management,

mainly focused on the United States: running
for congress (Black, 1972; Fox and Lawless, 2005).
4
Literature

Decision of running: number of available seats,
term limit requirements, party support, age, the
prospect of a better position,…

Potential candidates are more likely to seek office
when political conditions are more favourable
(Schlesinger, 1966).

Personality effects are relevant in explaining vote
and popularity => different candidates provide
different electoral outcomes (Frey & Schneider, 1978;
Lanoue & Headrick, 1994).
5
Literature

Politicians try to maximize their chances of winning
elections => the perception of the electoral success
affects their strategies (Nordhaus, 1975; Hibbs, 1977).

Studies for the U.S. conclude that when the electoral
support decreases, the probability of seeking
reelection falls (Moore & Hibbing, 1998; Theriault, 1998).

Popularity erosion or costs of ruling: the number of
consecutive terms in office may reduce the chances
of reelection (Mueller, 1970; Veiga & Veiga, 2004, 2010).
6
Literature

The economy is another important dimension that
affects incumbent’s electoral fortune.

For Portugal, Veiga & Veiga (2004, 2010) find
evidence of a reward/punishment mechanism
related to the economic performance.

Good local economic performance should increase
Mayor’s general sense of efficacy as a candidate.

National and/or Local economic conditions? Both.
N: Remmer & Gélineau (2003); Belanger & Gélineau (2004).
L: Squire & Fastnow (1994); Atkeson & Partin (1995), Hansen (1999).
7
Portuguese Municipal Elections

First municipal elections: December 1976.

In each municipality there is a Mayor, Town/City
Council (executive power) and a Municipal
Assembly (deliberative power).

The Mayor is the candidate that achieves the highest
number of votes (serves a 4-year term).

Mayor’s tasks: manage human resources, contracts,
licenses, projects and their timetable; territory organization,
social and economic development; supply local public goods
(water, transportation, housing, healthcare, education, culture). 8
Portuguese Municipal Elections

So, Mayors are responsible for the well-being of the
population that lives in the municipality => Mayors’
electoral fortune is tied to the economic environment.

Until the 2005 Municipal elections, there was no
legal limit on the number of the terms in office.

But a 3-term limit was established in 2005.

Thus, the period 1976-2005 represents an interesting
and unique case for studying the reasons why Mayors
run for another term in office.
9
Data

278 mainland municipalities.

Period: 1979-2005 (8 electoral periods).

Dependent variable (Recand) = 1 when a
Mayor is running for another term in office.

Several economic, political, and individual
variables are used as regressors,

controlled for at 3 levels: National, Regional
(NUTS III level) and Local (Municipal level).
10
Data – independent variables
National
UR
GDP
Inflation
Regional
Percentage change in the national unemployment rate+
UR
GDP_pc
Local
Percentage change in the regional (NUTS) unemployment rate+
PPI
Change in the Municipal Purchasing Power Index (in percentage)+
Wages
Change in Municipal real wages (in percentage)+
Employ
TotExpd
CapExpd
Change in Municipal employment (in percentage)+
Real GDP growth rate+
National inflation rate (percentage change in CPI)+
Real regional (NUTS) GDP per capita (thousands Euros)+
Real total expenditure per capita (Municipal – in thousands of Euros)
Real capital expenditure per capita (Municipal – in thousands of Euros)
Notes: + percentage change from the previous year and average percentage change over term.
11
Data – independent variables
Control
LnPop
SGov
TLGov
MajGov
Age
Male
Degree
Residence
Logarithm of the population living in each Portuguese municipality
Dummy variable = 1 if the local and national governments are of the
same party; 0 otherwise
Time in local government, i.e. number of terms in office
Dummy variable = 1 if the Mayor’s party has a majority; 0 otherwise.
Age of the Mayor running for another term
Dummy variable = 1 if the Mayor is male; 0 otherwise
Dummy variable = 1 if the Mayor has a university degree; 0 otherwise
Dummy variable = 1 if the Mayor lives in the municipality where he or
she was elected; 0 otherwise
Sources: Main Economic Indicators (OECD); IMF International Financial Statistics (IMF); Portuguese
National Institute of Statistics (INE); Marktest; Finanças Municipais (Municipal Finances DGAL); Quadros de Pessoal (database of the Portuguese Ministry of Labour); Technical Staff
for Matters Concerning the Electoral Process (STAPE).
12
Model

Probit model

estimated over a panel of municipalities and
elections.
Prob(recand  1 | x)  (x' β)
y  x it ' β   it ,
i  1,...,n, t  1,...,Ti ,
*
it
recandit  1 if y  0, and 0 otherwise.
*
it
13
Empirical Results
Table 1. Regressions with national and regional variables: Pooled probit (Average Marginal Effects)
Percentage changes from the previous year
National
Regional
(1)
(2)
(3)
(4)
UR
-0.0004
(-0.49)
GDP
0.0133
(1.15)
SGov
TLGov
MajGov
No. Obs.
0.0040
(1.11)
-0.0073
(-1.59)
0.0024
(0.49)
-0.0032**
(-2.56)
-0.0025**
(-2.03)
CapExpd
LnPop
0.0006
(0.66)
-0.0045
(-1.19)
GDP_pc
Inflation
Average percentage changes over term
National
Regional
(5)
(6)
(7)
(8)
-0.0109
(-1.27)
-0.0441**
(-2.50)
-0.0177***
(-3.69)
0.0888***
(4.77)
2,115
-0.0108
(-1.25)
-0.0429**
(-2.47)
-0.0179***
(-3.73)
0.0889***
(4.78)
2,115
0.0027
(0.55)
0.0005
(0.45)
0.0003***
(3.59)
0.0349**
(2.36)
-0.0474**
(-1.97)
-0.0255***
(-5.40)
0.0609**
(2.16)
1,103
0.0003***
(3.31)
0.0308**
(2.04)
-0.0450*
(-1.85)
-0.0257***
(-5.40)
0.0627**
(2.26)
1,103
-0.0088
(-1.05)
-0.0510***
(-2.75)
-0.0201***
(-4.48)
0.0862***
(4.41)
1,872
0.0001
(0.10)
-0.0089
(-1.06)
-0.0500***
(-2.71)
-0.0205***
(-4.57)
0.0854***
(4.39)
1,872
0.0003***
(3.36)
0.0256
(1.43)
-0.0438
(-1.60)
-0.0201***
(-3.81)
0.0329
(0.93)
828
0.0003***
(3.29)
0.0303**
(1.99)
-0.0449*
(-1.85)
-0.0257***
(-5.41)
0.0628**
14
(2.27)
1,103
Empirical Results: National & Regional

National and regional economic environment are not
very relevant for the Mayors’ decision of seeking
reelection (except inflation => negative effect).

The likelihood of seeking reelection increases:

when capital expenditures increase (improve local
economy => sign of Mayor’s intention of re-running);

when Mayors have majority in the Council (support);

in municipalities with more population (visibility,
power, wages indexed to the number of voters).
15
Empirical Results: National & Regional

The likelihood of seeking reelection decreases:


when a Mayor belongs to the party that leads the
national government:

offered a better position in the national gov.;

negative spillover effects from the national gov.;
for each additional term (popularity erosion or
costs of ruling reduce the chances of reelection).
16
Table 2. Regressions with local variables: Pooled probit (Average Marginal Effects)
% changes from the previous year
PPI
Wages
(2)
(3)
(4)
(5)
(6)
0.0054***
(2.61)
0.0052***
(2.63)
-0.0029
(-0.84)
0.0054***
(2.68)
0.0097*
(1.90)
0.0092*
(1.85)
0.0132*
(1.69)
0.0091*
(1.72)
Employ
TotExpd
0.0003
(0.39)
0.0002***
(3.04)
CapExpd
LnPop
SGov
TLGov
MajGov
Age
Male
Degree
Residence
No. Obs.
Average % changes over term
(1)
0.0371*
(1.79)
-0.0395
(-1.44)
-0.0149***
(-2.68)
0.0399
(1.09)
-0.0073***
(-3.79)
-0.0992
(-1.21)
0.0241
(0.86)
-0.0164
(-0.333)
795
-0.0010
(-0.45)
0.0002***
(2.85)
0.0003***
(3.35)
0.0332*
(1.88)
-0.0437
(-1.58)
-0.0144***
(-2.61)
0.0323
(0.90)
-0.0074***
(-3.82)
-0.0969
(-1.20)
0.0251
(0.89)
-0.0129
(-0.26)
795
0.0003***
(3.28)
0.0314*
(1.76)
-0.0471*
(-1.69)
-0.0147***
(-2.63)
0.0355
(0.98)
-0.0067***
(-3.35)
-0.1007
(-1.23)
0.0272
(0.95)
-0.0197
(-0.39)
778
0.0407*
(1.82)
-0.0413
(-1.51)
-0.0143**
(-2.54)
0.0319
(0.87)
-0.0070***
(-3.61)
-0.1026
(-1.26)
0.0221
(0.79)
-0.0108
(-0.22)
792
0.0003***
(3.32)
0.0327*
(1.73)
-0.0466*
(-1.69)
-0.0146***
(-2.63)
0.0293
(0.80)
-0.0070***
(-3.63)
-0.1051
(-1.28)
0.0243
(0.88)
-0.0063
(-0.13)
792
0.0003***
(3.26)
0.0329*
(1.75)
-0.0498*
(-1.79)
-0.0145***
(-2.57)
0.0318
(0.85)
-0.0064***
(-3.22)
-0.1025
(-1.26)
0.0258
(0.91)
-0.0143
17
(-0.29)
775
Empirical Results: Local

Local economic conditions (purch. power; expend) are
more important for Mayor’s decision than national.


Improvements in municipal economy => increases the
chances of reelection => incentive for re-running.
Political variables: only the coeff on TLGov remains
significant and with the expected sign (less observations).

Individual characteristics: only the Age of the
candidate is relevant (older Mayors have a lower
propensity to run for another term).
18
Conclusions

Local economic conditions matter more for Mayor’s
decision of re-running than national or regional ones.

Mayors try to influence local economic conditions
by increasing expenditures before elections to get
more electoral support in the next election.

Reelection choices are significantly related to the
expectations regarding the electoral results (in line
with traditional vote and popularity functions).

Mayors’ relevant economic time horizon is not
restricted to the recent past.
19
The End
Comments are welcome!
Thank you!
20